You likely know about Wired article, where they discovered from the video metadata that it was edited and that it actually used two source clips. You also know about the missing minute from 11:58:58 to 12:00:00. Another redditor used a different exiftool command for more metadata. There you can find the lengths of the two source video from the PantryDurationValue field, giving values 15556150ms and 23760550ms, which become 4:19:16 and 6:36:00 and together they sum to 10:55:16. The actual video is 10:52:24, so 2 minutes and 52 second are missing.
Also conveniently, the footage starts at 7:40:00 PM (4:20:00 to midnight, 40:18:58 if counting the missing minute ) and ends at 6:40:00 AM. So if we go by the theory that the 4:19:16 section was from 8/9/2019 and 6:36:00 section was from 8/10/2019 (explains the aspect ratio change), there are actually 18 seconds missing from 8/9/2019 and 4 minutes of extra footage compared to the source clip.
Also, if you look at PantryTracksMarkersStartTime field, there is a list of integers. They are likely frames where the editor added some comments or made edits. You can convert them to seconds by diving them by 60 (original fps). These seem to be timestamps for when people appear close to camera: ‘0:06:45’, ‘0:16:35’, ‘0:35:49’, ‘1:26:43’, ‘1:30:59’, ‘1:34:56’, ‘1:42:16’, ‘1:51:22’, ‘4:17:58’, ‘4:22:12’, ‘4:22:12’, ‘7:31:47′, ’10:01:15′, ’10:02:32′, ’10:02:59’, ‘0:06:45’, ‘0:16:35’, ‘0:35:49’, ‘1:26:43’, ‘1:30:59’, ‘1:34:56’, ‘1:42:16’, ‘1:51:22’, ‘4:17:58’, ‘4:22:12’, ‘4:22:12’, ‘7:31:47′, ’10:01:15′, ’10:02:32′, ’10:02:59’
EDIT:
I figured out the edit history time format. It seems that the missing minutes comes from cut to 4:19:16 segment, only 4:16:23.6 was used, which is the 2min52s difference.